6

I see it is possible to see all tags that exist, in order of most popular (most questions asked, presumably) to least. I quite enjoy contributing on stack overflow, but I find that some tags, which I would like to contribute to, have so many questions, I can't handle (like C++ tag). Which led me to wonder, if there is a way to find, for a given tag, which secondary tags have lowest answer ratio. For example, say I see the list of tags:

C# 60000
C++ 50000
python 40000
Java 30000

I'd like to know which tag that appears with C++ has smallest ratio of questions answered. So if I clicked on C++, I might see:

c++ 50000 
- multithread   4000 50%
- memory        3000 65%
- string        2000 35%

which would indicate that for questions that have the C++ tag, there are 4000 that also have the multithread tag, 3000 the memory tag, and 2000 the string tag; and that 65% of C++ questions that have the memory tag have an accepted answer, but only 35% of C++ questions that have a string tag have accepted answer.

This would help me choose what tag groups to monitor: the ones with lowest acceptance ratio, presumably the ones that are most in need of help.

Note that this % acceptance ratio could be shown with any combination of tags: just c++, c++ and multithread, c++ and multithread and memory, etc. The order of tags doesn't matter: ultimately it's the % acceptance of all questions asked for a selected group of tags, the drill-down just helps people find what they are most interested in answering.

If there is interest I'd be happy to implement such capability.

4
  • 2
    You might find the answer to your question by poking Data.SE a bit. There, you can run queries against the entirety of Stack Overflow's database (with some exceptions of personal information). data.stackexchange.com/stackoverflow/query/12473/… might be a query to start your explorations from.
    – user213963
    Commented Dec 13, 2013 at 4:30
  • @MichaelT if that were an answer, I would upvote it.
    – djechlin
    Commented Dec 13, 2013 at 4:45
  • @MichaelT very interesting, I'll give a shot at customizing the query
    – Oliver
    Commented Dec 13, 2013 at 13:39
  • @djechlin same here, and I'd accept it. Feel free to upvote the question ;)
    – Oliver
    Commented Dec 13, 2013 at 13:39

1 Answer 1

4

Much of Stack Exchange's data is available for queries via Data.StackExchange. You can write queries that do many things.

In this case, there's a rather old, existing query (yea, Data.SE search box is quite literal in what it searches for - they can be hard to find).

Least answered tags (relative)

This query returns back the tag, the number of unanswered questions, the number of answered questions, and the ratio between the two.

As of this time, the results look something like this:

Tag                      Unanswered Answered U/A Ratio      
------------------------ ---------- -------- -------------- 
facebook-likebox         17         199      0.085427135678 
facebook-comments        30         357      0.084033613445 
mediarecorder            15         185      0.081081081081 
mediaelement.js          30         395      0.075949367088 
avassetwriter            11         148      0.074324324324 
google-street-view       11         161      0.068322981366 
iscroll                  11         162      0.067901234567 
woocommerce              37         552      0.067028985507 
cisco                    15         238      0.063025210084 
pinterest                13         208      0.0625         
sharepoint2013           27         438      0.061643835616 
iis-8                    11         196      0.056122448979 
facebook-login           27         493      0.054766734279 
pdfbox                   18         350      0.051428571428 
slidingmenu              11         220      0.05           

If you decide to dig into this and fork the query, do check out the information on the magic links found in help. You will find if you change that Tag column name to [TagName] or something of that nature, it will become a link to the tag on the site.

2
  • Are you classifying "unanswered" as questions without answers? That would differ from how those are classified on SO proper. If I go to [facebook-likebox] for example, I see 207 "unanswered" questions: stackoverflow.com/questions/tagged/… out of 335 total. If I remember correctly, "answered" is attributed to questions with an upvoted or accepted answer only.
    – Brad Larson Mod
    Commented Dec 13, 2013 at 16:36
  • @BradLarson I didn't write the query, so I'd have to dig into it to see what / how it does what it does. Its a starting point for investigation into the question being asked. Its also Data.SE which isn't realtime data. So it may have been the right data a month ago or so, but not today. That said, looking into the code: SUM(CASE WHEN Post.AnswerCount = 0 THEN 1 ELSE 0 END) AS UnansweredCount seems to indicate that it is using the number of answers on the question being zero or not zero... upvotes and accepts don't factor into it.
    – user213963
    Commented Dec 13, 2013 at 22:33

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .